We promised full transparency when we started documenting the bot in public — and full transparency includes publishing small numbers, mixed results and one broker where the bot is, for now, in the red. This is the report of the forward test on a DEMO account of our Python bot, running at three brokers (IQ Option, Quotex and Deriv) with the final configuration applied around 02-07/07. No profit screenshots, no cherry-picking: the data as it stands.
Methodology (read this before the numbers)
The strategy came from a study with 45 days of real M1 candles from Deriv (official API), with a train/test split and a fresh out-of-sample window the calibration never saw. Reference: with an 87% payout, break-even is a 53.5% win rate. And the usual mantra: win rate is not profit — payout, time of day and execution decide the financial result. All on a DEMO account.
How we measure: executed trades vs. shadow mode
The bot records two different things, and confusing them is the classic recipe for fooling yourself:
(a) Executed trades — real orders sent to the broker, with time-of-day filters, minimum payout and risk management. It’s what matters in the end, but generates few samples per day.
(b) Shadow mode — on every candle, the bot checks whether the strategy’s signal would have won, even when it isn’t trading (outside trading hours, low payout, daily target hit). The n grows much faster and measures the quality of the signal on that specific feed, without execution noise.
(a) Executed trades — the raw numbers
| Broker | W / L | Win rate | n |
|---|---|---|---|
| IQ Option | 22 / 13 | 63% | 35 |
| Quotex | 18 / 19 | 49% | 37 |
| Deriv | — | — | 2 |
Mandatory context before drawing conclusions: Quotex’s 37 trades include operations prior to the final configuration and the atypical war day on 15/06 — a market regime no reversal strategy likes. Cutting to July only (already with the final config), Quotex sits at 8W/7L = 53%. Deriv has only 2 trades because the bot has just been ported there — n=2 is not data, it’s anecdote. And IQ Option’s 63% at n=35? Also too early. With samples this size, luck still shouts louder than statistics.
(b) Shadow mode — the signal on each feed
| Broker (feed) | Signal win rate | n |
|---|---|---|
| Deriv | 59.0% | 383 |
| IQ Option | 56.5% | 414 |
| Quotex | 53.1% | 224 |
The honest reading, in three points:
1. Shadow matches the study. The backtest projected a 57-60% win rate (see the full rules of the two strategies). Deriv (59.0%) and IQ (56.5%) are in or near that range, with n in the hundreds. This still isn’t proof — but it’s the behavior we would expect to see if the edge were real.
2. The n of executed trades is far too small for any conclusion. 35 trades at 63% and 37 at 49% are compatible with the same strategy running under different conditions — or with pure variance.
3. Quotex’s feed is “harder” than Deriv’s. Same strategy, same period, lower win rate (53.1% vs. 59.0%). The less exciting and more likely explanation: each broker has its own price feed. A strategy calibrated on Deriv candles is not automatically transferable to Quotex’s feed — and the data is showing exactly that. It’s the same reason we treat OTC feeds as their own animal.
Operational discovery: Quotex closes forex at night
Running the bot every day, we found out in practice: Quotex closes the real forex pairs around 18h BRT (UTC-3). That means the evening window (18h-23h59 BRT — i.e. 21h-2h UTC) — which was precisely the best one in our study — is not playable on Quotex. There, only the morning window counts. That’s why the evening shift went to Deriv and IQ Option. It’s the kind of detail no backtest reveals and that changes the architecture of the operation. Our data on the best hours and days for forex M1 goes deeper.
Risk management: what the 25 combos taught us
Before the forward test, we simulated a grid of 25 combinations of daily target and stop. Two conclusions paid for the effort:
A stop that’s too tight cuts off the recovery. Days that would have ended positive die at the early stop. No stop at all is even worse: the loss tail becomes unlimited — rejected without ceremony. The balance we adopted:
Daily target: 3× the stake — hit it, stop
Daily stop: 5× the stake — hit it, stop
Stake: fixed, NO martingale
Level cap: 10 (hard exposure lock)
On martingale: we don’t use it, we won’t use it, and be suspicious of any bot that does. Doubling the bet after losing creates no edge — it only transfers the risk to a bigger future catastrophe. If the strategy doesn’t survive with a fixed stake, it doesn’t survive; martingale just hides that for a while.
What this report does NOT say
It’s worth spelling out, because the internet will read whatever it wants: this is not proof that the bot is profitable. It’s a demo account — no emotional slippage, no real risk, and demo execution can differ from real. The trade n is small. IQ’s 63% may regress; Quotex’s 49% may improve. What we have is a signal consistent with the study at the two brokers with the most shadow data, and a clear warning that Quotex’s feed requires its own recalibration. We keep collecting and will publish the next reading — including if the numbers get worse.
Want to follow along or replicate the test? Deriv is the only one of the three with an official API — and the best feed for our strategy so far.
Open a Deriv demo account →Affiliate link. Always start — and stay a long while — on demo.
Read next
→ Double Reversal & Band Exhaustion: full rules of the 2 strategies
→ Is a 70% win rate a myth? The real ceiling of strategies
→ The bot project: Quotex, IQ Option and Deriv in Python
→ Deriv Bot: how to create trading bots without coding
→ check_win freezing on IQ Option? How we fixed it
Dan Machado
Developer of the IA Trader Pro project. Builds and documents in public a binary options bot in Python, tested on demo accounts at three brokers (Deriv, IQ Option and Quotex) — publishing the numbers that work and, above all, the ones that don’t.
Disclaimer: binary options are extremely high-risk products and most retail traders lose money. All results in this article are from a DEMO account, with a small sample — demo is not the same as a real account and nothing here is a promise of profit. Strictly educational content; not investment advice. Never trade with money you cannot afford to lose.
